I hope I don't sound too much like a rookie here, but what do you think is the best way to enter a position you're fundamentally bullish/bearish on? I trade primarily based off of macroeconomics so it isn't always clear when prices have reached their max/min before reversal... I think the dilemma I'm facing can best be explained by example. (Commissions are usually important but for the sake of this example let's assume there is no transaction cost.) Take a stock trading at 10, I want to enter long because I think it will go to 12 or I will be stopped out at 9. Here are 3 possible ways to enter:1. Just buy at 10, set the SL at 9 and take profit at 12.2. Buy some at 10, more at 9.90, more at 9.80, etc. This way, if I buy on the way down (but never taking on more than x% risk) I'm buying when the risk/reward is becoming increasingly favorable. The downside is if it heads right up to 12 I'm in a small position. Also, buying on the way down probably increases my odds of being stopped out at 9.3. Buy some at 10, more at 10.10, more at 10.20, etc. The advantage here is to buy as the price is moving in your favor, and if you take the initial position at 10 and it promptly drops to 9 you're in a very small position. The downside is you're buying as the risk/reward becomes less favorable. Depending exactly how you set it up, it may also be easier to get whipsawed. There are plenty of other ways to consider entering a position, but I think the above 3 examples explain the internal debate I've been having about this. What does everyone think is appropriate? Would anyone happen to know of a good book about this? I haven't found much in the way of trading literature so far and from what I have found there doesn't seem to be a good answer.

In my experience, the only way to figure this out is with simulation. It is very hard to make a simulation accurate. But it will help you think about the inputs which will dictate the correct approach.For example, consider a mean-reverting strategy. You believe prices will chop up and down, but longs and shorts will get screwed so that prices end up back where they started. Suppose the price starts rising above the average. Every dollar you sell short will be a dollar you won't have an opportunity to sell short with IF the price rises even higher before coming back down. So you are taking a 60% chance of making $2 and losing a 50% chance of a 60% chance of making $3. But you may also be taking a 50% chance of going down $1 before making $2, and passing up a chance to make $3 with only a 25% chance of losing $1 first. So if you feel safer to take more leverage if you wait for the $3, you may be losing a 50% chance of a 60% chance of making $5.It is very difficult to balance all these things in your head. So you need to make a simulation that somewhat represents the possible paths of the price, and their probabilities. A simulation should include at least four paths, and the probability of each path unfolding. Then the computer will see how each strategy fares in profit and variance.Suppose you are buying a stock because you expect earnings surprise. You will want to buy right before the earnings come out, to minimize pointless variance. But after, you may have alternate paths where the stock jumps then go sideways, the stock jumps then goes up a little more over the next few hours, or the stock jumps, and is very volatile but with a small upside bias after the initial jump. So how do you figure out how long to hold the stock after the jump, depending on the probability of different amounts of randomness and the probability of different amounts of bias? For me, the simplest approach is to do a simulation.

2. and 3. represent two very different styles of trading, the "mean reversion" and the "trend following" styles. Neither is right or wrong; in an efficient market they both have zero expected return but very different probability distribution of P&L (roughly speaking the meanrev has negative skewness and the trendfllwg has positive skewness). In real life the expected return is probably holding time and market specific; you would have to analyse a lot of data to determine which is better, as Farmer suggested, and it depends on the strategy (i.e. what "anomaly" are you triyng to capture). For what it is worth my impression is that short term equity traders follow the mean reversion style while currency traders are more often trendfollowing. Until you come to your own conclusion you could keep it simple and use method 1.

QuoteOriginally posted by: farmerIn my experience, the only way to figure this out is with simulation. It is very hard to make a simulation accurate. But it will help you think about the inputs which will dictate the correct approach.For example, consider a mean-reverting strategy. You believe prices will chop up and down, but longs and shorts will get screwed so that prices end up back where they started. Suppose the price starts rising above the average. Every dollar you sell short will be a dollar you won't have an opportunity to sell short with IF the price rises even higher before coming back down. So you are taking a 60% chance of making $2 and losing a 50% chance of a 60% chance of making $3. But you may also be taking a 50% chance of going down $1 before making $2, and passing up a chance to make $3 with only a 25% chance of losing $1 first. So if you feel safer to take more leverage if you wait for the $3, you may be losing a 50% chance of a 60% chance of making $5.It is very difficult to balance all these things in your head. So you need to make a simulation that somewhat represents the possible paths of the price, and their probabilities. A simulation should include at least four paths, and the probability of each path unfolding. Then the computer will see how each strategy fares in profit and variance.Suppose you are buying a stock because you expect earnings surprise. You will want to buy right before the earnings come out, to minimize pointless variance. But after, you may have alternate paths where the stock jumps then go sideways, the stock jumps then goes up a little more over the next few hours, or the stock jumps, and is very volatile but with a small upside bias after the initial jump. So how do you figure out how long to hold the stock after the jump, depending on the probability of different amounts of randomness and the probability of different amounts of bias? For me, the simplest approach is to do a simulation.Thanks Farmer, that is a great idea to run a computer simulation with at least 4 scenarios. That still sounds very complicated, I assume you'd have to be very careful making your scenarios... I mean, if the stock were at $10 you could have a scenario where it gaps down to .01 but the odds of that are very slim. I guess it comes down to just doing your best to make the right scenarios to approximate the most likely outcomes in accordance to the right probability distribution associated with estimated future volatility... Who ever said trading was easy? lolQuoteOriginally posted by: acastaldo2. and 3. represent two very different styles of trading, the "mean reversion" and the "trend following" styles. Neither is right or wrong; in an efficient market they both have zero expected return but very different probability distribution of P&L (roughly speaking the meanrev has negative skewness and the trendfllwg has positive skewness). In real life the expected return is probably holding time and market specific; you would have to analyse a lot of data to determine which is better, as Farmer suggested, and it depends on the strategy (i.e. what "anomaly" are you triyng to capture). For what it is worth my impression is that short term equity traders follow the mean reversion style while currency traders are more often trendfollowing. Until you come to your own conclusion you could keep it simple and use method 1.This really made me think... Depending on the trade, my approach could be either mean reverting or trend following. I might be best off using different entry strategies depending on the trade, or perhaps even adopt characteristics of both strategies. Thanks acastaldo!

In my opinion, you should always scale into positions, whatever the trading style. When your view about the market direction is correct, you enter more slowly and so give up a little of the upside. But when you are wrong, you exit from a much smaller position. The net result is that your trading costs go down and that your sharpe ratio goes up. As for buying on "dips" or on "confirmation", in my opinion, you should keep buying as long as the market has not invalidated your trade idea (the same rationale applies to the positioning of the stop loss).

QuoteOriginally posted by: QuasiRandomIn my opinion, you should always scale into positions, whatever the trading style. When your view about the market direction is correct, you enter more slowly and so give up a little of the upside. But when you are wrong, you exit from a much smaller position. The net result is that your trading costs go down and that your sharpe ratio goes up. As for buying on "dips" or on "confirmation", in my opinion, you should keep buying as long as the market has not invalidated your trade idea (the same rationale applies to the positioning of the stop loss).We could propose some sort of elegant theorem: Information always arrives gradually. As long as the smallest information grain is smaller than your smallest trading unit, you should never enter more than one unit at time.There may seem to be exceptions to this theorem, such as if an important economic number comes in really big. But the information still arrives in stages. Your program parses it in stages. It comes over the network byte-by-byte. It is only because your method of consuming the information is too large-grained, or your minimum trade unit is too large-grained, that it appears there is an exception.So really you should be entering 1 unit when you get the sign of the number, or the first digit over the network, and another unit when you get the next digit, and another unit when your entire understanding of the transmission is verified to a high degree of confidence. If you cannot do this, the problem is not with the theorem, it is with your systems.The theorem is invalid if there truly are threshhold moments, where it suddenly becomes correct to place a limit order of your maximum position size, all at once. I don't doubt there are such threshhold moments. The theorem is probably wrong, I was just proposing it.

woden - there is no hard and fast rule for what is better. "trading" is a lengthy process, consisting of elements ranging from risk budgeting to the actual trade execution. it's helpful to separate the former from the latter. risk budgeting => how much of my total capital do I allocate to a particular trade? this typically requires an assumption about your subjective probability about future events relative to where the market is pricing those probabilities. a good place to start is the Kelly Criterion. trade execution => given a risk budget per trade you can determine combinations of stop loss (levels), position size, which then map into possible entry points that fufill those combinations. that's the science, making it work well is the art.

QuoteOriginally posted by: QuasiRandomIn my opinion, you should always scale into positions, whatever the trading style. When your view about the market direction is correct, you enter more slowly and so give up a little of the upside. But when you are wrong, you exit from a much smaller position. The net result is that your trading costs go down and that your sharpe ratio goes up. As for buying on "dips" or on "confirmation", in my opinion, you should keep buying as long as the market has not invalidated your trade idea (the same rationale applies to the positioning of the stop loss).This is not supported in back tests. If size allows, putting a position on without scaling is preferred. Scaling out is a different story.

Depends on several factors such as:- Whether its a long/medium/short term position- Long or short position. More risky for short positions (no limit of upside) vs long (cannot go below 0)- Style of trading (mean reversion and against the trend vs. trend following)- I prefer scaling into short/medium term positions in direction of the trend because when you lose, you lose little, when you win, you win a bit more because the price confirms and re-confirms your view in the direction of your trade.- For long term positive carry position, I prefer stake building via scaling into the position at dips. The positive carry helps in offsetting the transaction cost and also due to longer duration it contributes in bringing your average price closer to the market price.